Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Gastroenterol. Aug 28, 2026; 32(32): 120382
Published online Aug 28, 2026. doi: 10.3748/wjg.120382
Automatic recognition of tumour-infiltrating lymphocytes in pathological biopsy images of the gastric mucosa
Yu Fan, Su-Nan Wang, Bo Jiang, Ying-Ying Li, Chao-Ya Zhu, Xing-Hai Liao, Fa-Shun Zhang, Yang-Kun Wang
Yu Fan, Department of Pathology, Shaanxi Provincial Hospital of Traditional Chinese Medicine, Xi’an 710028, Shaanxi Province, China
Su-Nan Wang, Ying-Ying Li, Shenzhen Polytechnic University, Shenzhen 518055, Guangdong Province, China
Bo Jiang, Department of Pathology, People’s Liberation Army Joint Logistic Support Force 990th Hospital, Zhumadian 463000, Henan Province, China
Chao-Ya Zhu, Department of Pathology, Third Affiliated Hospital, Zhengzhou University, Zhengzhou 450052, Henan Province, China
Xing-Hai Liao, Department of Surgery, Southern Medical University Shenzhen Hospital, Shenzhen 518110, Guangdong Province, China
Fa-Shun Zhang, Department of Pathology, Xuchang Central Hospital, Xuchang 461099, Henan Province, China
Yang-Kun Wang, Department of Pathology, The Fourth People’s Hospital of Longgang District, Shenzhen 518123, Guangdong Province, China
Co-first authors: Yu Fan and Su-Nan Wang.
Author contributions: Wang YK conceived and designed the study; Fan Y ,Liao XH and Jiang B collected data, sorted pathological samples, and drafted the manuscript; Wang SN and Fan Y constructed the model, optimized the algorithm, implemented ablation experiments, and analyzed data; Jiang B and Fan Y performed pathological image annotation, established ground truth, and verified interobserver consistency; Li YY and Fan Y conducted image preprocessing including regions of interest extraction and colour deconvolution, and constructed the dataset; Zhu CY, Fan Y and Zhang FS collated clinical follow-up data and performed survival analysis of early gastric cancer patients; Liao XH , Fan Y and Zhang FS collected multicentre samples and confirmed clinical information; Zhang FS and Fan Y performed pathological diagnosis of gastric mucosal lesions and conducted double-blind verification of sample types; Wang YK acquired funding, revised the manuscript, and gave final approval of the version to be published. Fan Y and Wang SN contributed equally to this work as co-first authors.
Supported by Shenzhen Basic Research Special Natural Science Foundation Project, No. JCYJ202506044185911015.
Institutional review board statement: Approval from the Ethics Committee of Shaanxi Provincial Academy of Traditional Chinese Medicine and Shaanxi Provincial Hospital of Traditional Chinese Medicine, Exemption Approval No. (2023) Lunshenmian No. (15), date: January 10, 2023. The study complied with the Declaration of Helsinki and China’s Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects.
Informed consent statement: Written informed consent was obtained from all participants.
Conflict-of-interest statement: The authors declare no conflicts of interest.
Data sharing statement: All data generated or analyzed during this study are included in this published article.
Corresponding author: Yang-Kun Wang, Department of Pathology, The Fourth People’s Hospital of Longgang District, No. 2 Jinjian Road, Nanwan Subdistrict, Longgang District, Shenzhen 518123, Guangdong Province, China.
dr.wyk@163.com
Received: February 26, 2026
Revised: March 12, 2026
Accepted: April 21, 2026
Published online: August 28, 2026
Processing time: 159 Days and 11.7 Hours
BACKGROUND
The immune landscape of the tumour microenvironment, particularly tumour-infiltrating lymphocytes (TILs), is pivotal in the progression of gastric mucosal lesions. However, conventional manual assessment of TILs is hindered by interobserver variability, poor reproducibility, and labour-intensive quantification, restricting its clinical utility.
AIM
To develop an artificial intelligence (AI)-based metric, the gastric-AI-TIL (G-AI-TIL) index, for precise lesion grading and prognostic stratification in early gastric cancer (EGC).
METHODS
We retrospectively collected 320 whole-slide images of gastric mucosal biopsy samples from three medical centres. A multiscale two-stage convolutional neural network (CNN), comprising a gastric-CNN for lesion segmentation and a G-TIL-CNN for TIL enumeration, was constructed. Image preprocessing involved region of interest extraction and Ruifrok-Johnston colour deconvolution to normalize staining variations. Two senior pathologists performed the annotations in a double-blind manner to establish the ground truth. The prognostic value of the G-AI-TIL was evaluated using Kaplan-Meier and multivariate Cox regression analyses.
RESULTS
In the independent test set (n = 96), the G-TIL-CNN model achieved an accuracy of 99.2% (kappa = 0.98), significantly outperforming manual evaluation by pathologists (86.3% accuracy; P < 0.001). A progressive increase in the G-AI-TIL index correlated with lesion malignancy, as follows: Chronic atrophic gastritis < intestinal metaplasia < high-grade intraepithelial neoplasia < EGC (P < 0.001). Furthermore, a high G-AI-TIL (≥ 28.5%) was identified as an independent protective factor for both 3-year disease-free survival [hazard ratio (HR) = 0.58] and overall survival (HR = 0.55; P = 0.009) in patients with EGC.
CONCLUSION
The proposed AI model enables objective, high-throughput quantification of TILs. The G-AI-TIL index serves as a robust biomarker for grading gastric mucosal lesions and stratifying EGC risk, providing a quantitative basis for personalized treatment strategies, such as the decision to perform endoscopic resection and surgery.
Core Tip: This study retrospectively collected 320 whole-slide images of gastric mucosal biopsies and constructed a two-stage convolutional neural network model, which underwent multi-step preprocessing and annotated training. The verification results showed the model achieved a 99.2% accuracy in tumour-infiltrating lymphocytes (TIL) identification. It was found that the gastric-artificial intelligence-TIL (G-AI-TIL) index increases with the elevated malignancy of gastric mucosal lesions, and a high G-AI-TIL index acts as an independent protective factor for the survival of early gastric cancer patients. This research provides a novel objective tool for the precise diagnosis and treatment of gastric cancer.
